Accuracy and reproducibility of absolute nannoplankton abundances using the filtration technique in combination with a rotary sample splitter

The filtration technique is a widely applied method to calculate absolute coccolith abundances in paleoceanographic, paleoclimatic and ecological studies. Here, we have tested the accuracy and reproducibility of absolute coccolith abundance estimates obtained with the filtration method in combination with a rotary sample divider. The examination of two different filtration devices, a funnel and an in-line system, analysing a known number of microbeads per weight, revealed that counts obtained with funnel systems are up to Ø 18.6% too high. Repeated analyses (15×) of a fine-fraction sediment sample revealed a standard deviation of ±11.4% for the in-line system and ±15.2% for the funnel system, respectively. The comparison between three sub-samples (three splits of 1/30) showed a standard deviation of 4.6% for the in-line system and 9.9% for the funnel system, respectively. Therefore, we assume that the potential split error for the rotary splitter is ≤4.6%. mparison of our results with estimates for the same samples analysed with the random settling and the “spiking with microbeads and spraying method” (SMS method) reveal that only the data obtained with the SMS and in-line filter method are statistically equal. The results obtained with the settling method are too low, whilst the results obtained with funnel filtration are too high in terms of coccolith absolute abundances. Our results confirm that the filtration method, including sample splitting with a rotary splitter, is a suitable method for the analysis of absolute coccolith abundances. However, the use of different filtration devices appears to affect the accuracy and reproducibility of the results. Furthermore, the number of splitting steps should be as small as possible to avoid an increase of error due to error propagation. In order to increase the reliability of the filtration method, we suggest calibrating a filtration device with microbeads as a standard measure. The reported reproducibility of the various preparation techniques may help to compare data sets obtained with different methods.